It is a Ximple (http://goreliance.com/devel )data structure to hold XML
tree.. It goes like this...
<ximple_tree> ::= [ <ximple_tag> ... ]
<ximple_tag> ::= { tag_name => <tag-name>,
attribs => <attribs>
content => <ximple_tree>
When we create XML from say comma delimited or fixed width records, of
which there could be thousands, we end up with a side span that contains
thoudands of children under the root node. It is this that breaks
Any2XML. There would be less problem with fewer children, as everyone
agrees :)
One solution is to print out the children every say x number of records.
This is what we have done. This needs a parameter from the user as
Any2XML itself does not know what it is parsing.
This is the code that creates the structure...
# -------------inside Any1xml -----------
sub xml_routine {
my ($tag,$text) = @_;
my @ret;
while (my @mem = get_match($tag,\$text)) {
push (@ret,{tag_name=>$tag->{tag_name},
content=>get_content($tag,\@mem),
attrib=>get_attrib($tag,\@mem),
tag_type=>($tag->{attrib}{'_fill'}||$tag->{attrib}{'%fill'})?undef:$tag-
>{tag_type}});
}
return \@ret;
}
>From the suggestions, I gather that by tieing this structure to an
array, we can print out the array as each element is created.
Questions: Could the "Tie" be done outside the module so Any2xml is not
aware of it? The array @ret is lexically scoped here. If it is globally
scoped (our @ret) this should be possible right?
---Closing information gaps-----
Ranga Nathan, Reliance Technololgy
>>Live demo at http://any2xml.com/docs/timesheet_demo.shtml<<
>>Get free COBOLExplorer at http://goreliance.com/download-products <<
> -----Original Message-----
> From: Brian Reichert [mailto:[EMAIL PROTECTED]]
> Sent: Monday, September 30, 2002 12:30 PM
> To: Ranga Nathan
> Cc: [EMAIL PROTECTED]
> Subject: Re: [Boston.pm] Large hashes
>
>
> On Sun, Sep 29, 2002 at 09:31:01PM -0400, Ranga Nathan wrote:
> > Are there any special techniques for handling large hashes. When
> > building an XML tree as a Perl data structure, at times
> more and more
> > memory is being consumed eventually runs out of memory. Other than
> > partitioning the problem is there any known technique?
>
> Would this simple be a matter using tie() with your hash? I
> suppose it depends on what data structures are in your hash.
>
> When you say you have a 'large hash', do you mean that you
> have lots of keys? Or do you mean that you're shoving lots
> of big key, values into it? Or do you mean that it contains
> references to really big data structures?
>
> > ---Closing information gaps-----
> > Ranga Nathan, Reliance Technololgy
> > >>Live demo at http://any2xml.com/docs/timesheet_demo.shtml<<
> > >>Get free COBOLExplorer at
> http://goreliance.com/download-products <<
> --
> Brian 'you
> Bastard' Reichert <[EMAIL PROTECTED]>
> 37 Crystal Ave. #303 Daytime number: (603) 434-6842
> Derry NH 03038-1713 USA Intel
> architecture: the left-hand path
>
_______________________________________________
Boston-pm mailing list
[EMAIL PROTECTED]
http://mail.pm.org/mailman/listinfo/boston-pm
